Oryzabase contains 16,000 biological resources, including wild rice, mutants, and experimental strains, together with 4000 trait genes collected from journal articles. The database also contains genetic and genomic maps, and 20,000 items fi-om the published literature. Of these items, the most valuable are the biological resources, particularly the wild rice and mutant collections, because they have been thoroughly characterized by researchers and are aecompanied by large quantities of data. The next most valuable content is the trait gene dictionary, which has been maintained continuously since 1995 and is based on published literature. Oryzabase also plays a role as an international site for the submission of rice genes. Besides maintaining original data, Oryzabase compiles genomic data that are systematically integrated with existing data holdings and provided with a user-friendly interface. Oryzabase can be accessed at http://www.shigen.nig.ac.jp/rice/oryzabase/.

Momilactones A and B were isolated as plant growth inhibitors from rice husks by Kato et al.1) In 1977, Cartwright et al.2) reported that momilactones A and B accumulated in rice leaves infected with rice blast fungus (Pyricularia oryzae) and were designated as the first phytoalexins from the Gramineae plants.

In addition to maintaining a rice gene list, Oryzabase, a database of rice science, has also been hosting a website to submit newly identified genes. The current rice gene dictionary is based on a traditional rice trait gene dictionary and extended through time with new genes which are manually retrieved from journal articles. Currently over 4000 genes are listed. However it is still insufficient in both quantity and quality. Recently we started to apply automated extraction before manual annotation to enhance the efficiency of gene extraction. To improve the quality of information, we started collecting relevant DNA accessions and LOCUS-ID. In Oryzabase, we try to employ the help of users to overcome this information insufficiency by adding a comment box in each gene page to allow researchers to write additional information. Besides adding GO (gene ontology) and TO (trait ontology) to the genes, PO (plant ontology) has been newly assigned to enable better semantic query processing. The current status of gene extraction and ontology assignment and its related problems will be introduced.

D4Oryzabase (http://www.shigen.nig.ac.jp/rice/oryzabase/) is a comprehensive rice science database [1]. It houses a variety of genetic resources, relevant literatures, gene dictionary, DNA sequences, and basic information such as developmental biology and anatomy. In order to keep the gene dictionary up-to-date, literature annotation has been conducted manually since 1995. However as the publication of journal articles increases year by year after genomic sequences were released, it became more difficult to update the dictionary timely and in high quality without sufficient annotators. To overcome this difficulty, we applied machine learning and text-mining to extract known and unknown genes from journals. The machine extraction followed by manual annotation achieved promising results and increased efficiency in manual annotation. Furthermore a direct submission system where rice researchers can deposit new genes according to the standardized nomenclature [2] became operational in 2008. Recent advances will be introduced. AbstractFeedback to a gene on-line gene submission system Rice Textpresso abstract

In addition to maintaining a rice gene list, Oryzabase, a database of rice science, has also been hosting a website to submit newly identified genes. The current rice gene dictionary is based on a traditional rice trait gene dictionary and extended through time with new genes which are manually retrieved from journal articles. Currently over 4000 genes are listed. However it is still insufficient in both quantity and quality.Recently we started to apply automated extraction before manual annotation to enhance the efficiency of gene extraction. To improve the quality of information, we started collecting relevant DNA accessions and LOCUS-ID. In Oryzabase, we try to employ the help of users to overcome this information insufficiency by adding a comment box in each gene page to allow researchers to write additional information. Besides adding GO (gene ontology) and TO (trait ontology) to the genes, PO (plant ontology) has been newly assigned to enable better semantic query processing. The current status of gene extraction and ontology assignment and its related problems will be introduced.